cpufreq(4) panic on RELENG_7 (was: Re: Call for bfe(4) testers.)

Tue Aug 12 08:36:30 UTC 2008

2008/8/11 John Baldwin <jhb at freebsd.org>:
> On Monday 11 August 2008 12:35:17 pm pluknet wrote:
>> 2008/8/11 John Baldwin <jhb at freebsd.org>:
>> > On Saturday 09 August 2008 07:16:37 am Ulrich Spoerlein wrote:
>> >> Hi John,
>> >>
>> >> I now figured out the "who", the "why" still eludes me.
>> >>
>> >> So, after your MFC of ichss.c on June 27th the device now attaches at my
>> >> laptop. It didn't before, so it could cause no trouble.
>> >>
>> >> With ichss loaded, the kernel will panic 1-3 minutes after powerd has
>> >> been started (if I kill powerd early enough, it seems pretty stable).
>> >>
>> >> I'm now running a kernel from 2008-08-08 with
>> >> hint.ichss.0.disabled="1"
>> >
>> > Ok.  Can you get a crashdump from a crash?
>> >
>>
>> ehm,. I am not Ulrich Spoerlein, but I can help with this issue.
>>
>> my crashdump from kgdb and some debug info.
>> (ouch, I forgot to include it in my prev. mail
>> http://lists.freebsd.org/pipermail/freebsd-stable/2008-August/044182.html )
>>
>> wbr,
>> pluknet
>>
>> Unread portion of the kernel message buffer:
>>
>>
>> Fatal trap 12: page fault while in kernel mode
>> fault virtual address   = 0x38
>> fault code              = supervisor read, page not present
>> instruction pointer     = 0x20:0xc056cf46
>> stack pointer           = 0x28:0xe6592ac8
>> frame pointer           = 0x28:0xe6592ac8
>> code segment            = base 0x0, limit 0xfffff, type 0x1b
>>                         = DPL 0, pres 1, def32 1, gran 1
>> processor eflags        = interrupt enabled, resume, IOPL = 0
>> current process         = 2507 (powerd)
>> Physical memory: 1014 MB
>> Dumping 120 MB: 105 89 73 57 41 25 9
>>
>> #0  doadump () at pcpu.h:195
>> 195     pcpu.h: No such file or directory.
>>         in pcpu.h
>> (kgdb) bt
>> #0  doadump () at pcpu.h:195
>> #1  0xc0458f89 in db_fncall (dummy1=-1010027648, dummy2=0, dummy3=0,
>>     dummy4=0xe6592860 "0╛йц") at /media/src-7/sys/ddb/db_command.c:516
>> #2  0xc045953a in db_command (last_cmdp=0xc07dcf14, cmd_table=0x0,
> dopager=1)
>>     at /media/src-7/sys/ddb/db_command.c:413
>> #3  0xc0459655 in db_command_loop ()
> at /media/src-7/sys/ddb/db_command.c:466
>> #4  0xc045b17c in db_trap (type=12, code=0)
>>     at /media/src-7/sys/ddb/db_main.c:228
>> #5  0xc0575023 in kdb_trap (type=12, code=0, tf=0xe6592a88)
>>     at /media/src-7/sys/kern/subr_kdb.c:524
>> #6  0xc07460bf in trap_fatal (frame=0xe6592a88, eva=56)
>>     at /media/src-7/sys/i386/i386/trap.c:890
>> #7  0xc074636b in trap_pfault (frame=0xe6592a88, usermode=0, eva=56)
>>     at /media/src-7/sys/i386/i386/trap.c:812
>> #8  0xc0746d36 in trap (frame=0xe6592a88)
>>     at /media/src-7/sys/i386/i386/trap.c:490
>> #9  0xc072fd4b in calltrap () at /media/src-7/sys/i386/i386/exception.s:139
>> #10 0xc056cf46 in device_is_attached (dev=0x0)
>>     at /media/src-7/sys/kern/subr_bus.c:2228
>> #11 0xc0512de7 in cf_set_method (dev=0xc3c9c880, level=0xc4525ef4,
>>     priority=100) at /media/src-7/sys/kern/kern_cpu.c:332
>> #12 0xc0511452 in cpufreq_curr_sysctl (oidp=0xc3c8bac0, arg1=0xc3bc7c00,
>>     arg2=0, req=0xe6592ba4) at cpufreq_if.h:32
>> ---Type <return> to continue, or q <return> to quit---
>> #13 0xc0554b67 in sysctl_root (oidp=Variable "oidp" is not available.
>> )
>>     at /media/src-7/sys/kern/kern_sysctl.c:1306
>> #14 0xc0554cd1 in userland_sysctl (td=0xc4245440, name=0xe6592c14,
> namelen=4,
>>     old=0x0, oldlenp=0x0, inkernel=0, new=0xbfbfe7c4, newlen=4,
>>     retval=0xe6592c10, flags=0) at /media/src-7/sys/kern/kern_sysctl.c:1401
>> #15 0xc0555a7c in __sysctl (td=0xc4245440, uap=0xe6592cfc)
>>     at /media/src-7/sys/kern/kern_sysctl.c:1336
>> #16 0xc07466d5 in syscall (frame=0xe6592d38)
>>     at /media/src-7/sys/i386/i386/trap.c:1035
>> #17 0xc072fdb0 in Xint0x80_syscall ()
>>     at /media/src-7/sys/i386/i386/exception.s:196
>> #18 0x00000033 in ?? ()
>> Previous frame inner to this frame (corrupt stack?)
>> (kgdb) f 11
>> #11 0xc0512de7 in cf_set_method (dev=0xc3c9c880, level=0xc4525ef4,
>>     priority=100) at /media/src-7/sys/kern/kern_cpu.c:332
>> 332                     if (!device_is_attached(set->dev)) {
>> (kgdb) list
>> 327             }
>> 328
>> 329             /* Next, set any/all relative frequencies via their drivers.
> */
>> 330             for (i = 0; i < level->rel_count; i++) {
>> 331                     set = &level->rel_set[i];
>> 332                     if (!device_is_attached(set->dev)) {
>> 333                             error = ENXIO;
>> 334                             goto out;
>> 335                     }
>> 336
>> (kgdb) p level.rel_count
>> $1 = 1986356271
>> (kgdb) p i
>> $2 = 0
>> (kgdb) p level.rel_set
>> $3 = {{freq = 0, volts = 0, power = 0, lat = 0, dev = 0x0, spec = {0, 0, 0,
>>       0}}, {freq = 0, volts = 0, power = 0, lat = 0, dev = 0x0, spec = {0,
> 0,
>>       0, 0}}, {freq = 0, volts = 0, power = 0, lat = 0, dev = 0x0, spec =
> {0,
>>       0, 0, 0}}, {freq = 0, volts = 0, power = 0, lat = 0, dev = 0x0, spec =
> {
>>       0, 0, 0, 0}}, {freq = 0, volts = 0, power = 0, lat = 0, dev = 0x0,
>>     spec = {0, 0, 0, 0}}, {freq = 0, volts = 0, power = 0, lat = 0,
>>     dev = 0x656e7552, spec = {828858701, 1162760014, 0, 134632492}}, {
>>     freq = 0, volts = 53, power = -1024, lat = -1, dev = 0x7fffffff, spec =
> {
>> ----- and so on-----
>>
>> Also there are very unusual (and high) numbers in sysctl
> dev.cpu.0.freq_levels.
>
> Which cpufreq drivers are you using?  Can you narrow down your panics (and
> weird frequencies in the sysctl) to being caused by a specific cpufreq
> driver?

They are est/p4tcc/ichss.
hint.ichss.0.disbled="1" helped me to avoid panics and those weired
freqs in dev.cpu.
Now it shows:
cpu0: <ACPI CPU> on acpi0
est0: <Enhanced SpeedStep Frequency Control> on cpu0
p4tcc0: <CPU Frequency Thermal Control> on cpu0
and dev.cpu.0.freq_levels are as expected (and as it was earlier
before this problem):
 1600/-1 1400/-1 1225/-1 1200/-1 1050/-1 1000/-1 875/-1 800/-1 700/-1
600/-1 525/-1 450/-1 375/-1 300/-1 225/-1 150/-1 75/-1

wbr,
pluknet